129 results about "Event graph" patented technology

The invention discloses a multi-modal power grid fault diagnosis method and system based on a causal event atlas, and belongs to the technical field of intelligent operation and maintenance of power systems. The method comprises the following steps: preprocessing historical fault case data of power grid equipment, and constructing a causal event atlas database; when a fault diagnosis request is received, analyzing the fault diagnosis request by the planning agent, generating an initial fault hypothesis set in combination with power field knowledge, endowing a corresponding credibility score to the fault hypothesis, retrieving the cause subgraph as evidence in an iterative loop, and updating the credibility score of the fault hypothesis by the reasoning agent; and generating a multi-modal diagnosis report after the termination condition is met. According to the method, the problems of insufficient causal modeling and poor interpretability of a traditional method are solved, and the diagnosis accuracy, efficiency and user credibility are remarkably improved.

The invention relates to the technical field of information, and provides a bulk commodity transaction system based on a block chain, and the system comprises a parameter and publishing module which uniformly receives and stores parameters; the unified examination and approval engine performs isomorphic arrangement and gray level updating through a configurable template and a state machine; consistency snapshots and recovery are paused / abnormally collected snapshots and monotonically replayed, and it is guaranteed that the sequence is consistent with a window; the security fund risk control uses DSL + fund DAG to drive freezing / unfreezing / deduction and account checking; and block chain evidence storage and auditing execute field-level hash and pedigree anchoring to realize verifiable traceability. According to the system, an end-to-end consistency and verifiability framework is constructed around the aspects of field-level fingerprinting, unified examination and approval, windowed stable sorting, consistency snapshot and monotone replay, fund event patterning, contract assembling and signature anchoring, settlement / final recalculation and pedigree uplink and offline verification; and under the conditions of abnormity, recovery and version evolution, each link keeps uniform caliber, stable sorting and approvable state.

The invention discloses a power grid dispatching report generation method based on natural language processing, relates to the technical field of power grid dispatching report processing, and aims at solving the problems that a traditional report generation mode is low in efficiency and large in subjective deviation. The method comprises the steps that multi-source data are integrated, the data are preprocessed, and multi-dimensional time sequence data are output; generating an event atlas based on the high-dimensional data abstract and the state recognition result; converting the structured data into basic natural language expression, and filling dynamic content; generating a report text; generating a dynamic visual chart and an image-text report based on the generated report text, the event graph and trend data generated by the time sequence analysis model; an image-text report is adopted, the report is divided into key suggestions and detail descriptions, and highlighted high-risk event prompts are generated. According to the method, the report generation efficiency can be improved, subjective deviation in manual analysis is eliminated, and objectivity and specialty of report content are ensured.

The invention discloses a field event element extraction method and system based on prompt learning and knowledge completion, and the method comprises the steps: converting an extraction task into a constrained generation problem through structured prompt, achieving the stable extraction under a small number of labeled samples through the existing knowledge prior of a large language model, reducing the dependence on large-scale labeled data, and improving the extraction efficiency. And the cross-domain adaptive capacity is improved. A vectorization knowledge retrieval mechanism is introduced, related evidences are obtained from an external knowledge source, elements which are not clearly expressed in a text are complemented, the integrity of information is enhanced, and the accuracy and credibility of a result are improved through knowledge verification. And through a self-adaptive fusion mechanism of prompt and knowledge, the model can flexibly balance text context and external knowledge, and the robustness of complex context and fuzzy expression is improved. And finally, standardized and structured element data are output, a convenient interface is provided for downstream event atlas construction and analysis tasks, and the automation degree of domain knowledge processing and the system integration efficiency are improved.

The invention discloses a smart city event perception and emergency scheduling method based on artificial intelligence. The method comprises the following steps: S1, constructing a standardized structured sample; s2, forming a city sub-event graph; s3, constructing a Bayesian event inference network, and synchronously recording intermediate layer feature response; s4, outputting a dynamic thermodynamic diagram as an event perception model; s5, constructing a multi-target scheduling graph, and generating an emergency scheduling task; s6, taking training of a lightweight student model as a target, constructing a fusion type knowledge distillation mechanism for edge deployment, and fusing teacher model output, intermediate layer feature response and Bayesian causal structure consistency as a distillation target; and S7, adjusting an event state association structure in the Bayesian event inference network and a response weight distribution mechanism of the lightweight student model. The method has the advantages of high reasoning precision, low response delay and edge deployment.

The invention provides a monocular depth estimation method and product based on convolution compensation dual-channel self-attention, and relates to the field of computer vision. Comprising the following steps: converting an event flow of a target scene into three-dimensional tensor representation; obtaining event image fusion multi-scale spatial features based on the image of the target scene and the three-dimensional tensor representation; modeling spatial context correlation in a spatial dimension by utilizing event image fusion multi-scale spatial features through a context modeling self-attention branch to obtain a context modeling self-attention result; through a modal fusion self-attention branch, using the event image to fuse the modal correlation of the multi-scale spatial features in the channel dimension modeling image and the event, and obtaining a modal fusion self-attention result; and pixel-level depth prediction is carried out by using a context modeling self-attention result and a modal fusion self-attention result to obtain a depth map, so that complementary characteristics between an event and an image are fully mined, fine-grained depth fusion expression is realized, and depth estimation precision and generalization ability are effectively improved.

The invention discloses an intelligent fire-fighting hidden danger identification method based on multi-modal fusion, and the method comprises the steps: collecting multi-source heterogeneous data in a building environment, and carrying out the unified processing of the data, and obtaining standardized multi-modal original input data; event anchor point detection is carried out on the multi-modal input, key events are extracted, and a corresponding multi-modal event sequence is generated; constructing an event graph containing time and causal edges, and outputting an alignment event stream by using a space-time coupling causal alignment module; importing BIM and air duct structure information to establish a spatial topological graph, and fusing event streams to generate spatial constraint features; constructing a self-evolution semantic decision map based on the fusion features, and dynamically adjusting node weights and edge connection relationships; and carrying out hidden danger identification and grade division on the real-time data by using the decision map, and outputting an early warning signal and positioning information. According to the method, high-precision identification and intelligent early warning of fire-fighting hidden dangers are realized through space-time causal alignment, spatial topology fusion and self-evolution semantic decision of multi-modal data.

The invention discloses a video event association reasoning method based on a knowledge graph. The method comprises the following steps: S1, collecting video data and preprocessing the video data; s2, constructing an event graph according to a preprocessing result, and generating a state representation sequence through a graph convolutional network; s3, constructing a DreamerV3 world model, inputting a state representation sequence, and generating an event trajectory set; s4, performing trajectory value evaluation operation on the event trajectory set, and constructing a target path based on an evaluation result; s5, executing a structure alignment operation in the event graph according to the target path, and generating a reasoning knowledge graph; and S6, based on the inference knowledge graph, outputting a video event association result according to a time sequence. According to the method, video identification, map modeling and trajectory prediction methods are fully fused, and an efficient closed loop of key event extraction and map-level causal reasoning is realized.

The invention discloses a digital work card-driven business window multi-source recording data fusion intelligent analysis system, and particularly relates to the technical field of voice processing. The system is based on asynchronous audio streams collected by a wearable digital work card, an array microphone and environment pickup equipment; structured alignment and unified time reference construction of multi-source recording data are realized by adopting the steps of anchor point detection, event graph construction, cross-source matching, elastic time distortion alignment and the like. The system completes cross-source event marking by introducing multiple types of anchor point detectors (semantic keywords, acoustic abrupt changes and prompt tones); constructing an event alignment graph and executing confidence classification in combination with a unified clipping and cross sliding matching strategy in the anchor point pairing process; and then performing multi-scale alignment and resampling on the audio stream based on a layered elastic time warping method, and finally outputting a frame-level synchronous voice data stream to provide unified time support for downstream speaker separation, behavior auditing and semantic mapping.

The invention discloses an automatic story generation method and system based on a user portrait, and relates to the technical field of user portrait modeling, and the method comprises the steps: collecting real-time interaction data of a user, generating a candidate keyword set in combination with an RAKE algorithm, calculating a VAD emotion vector, and generating an emotion resonance keyword set based on the candidate keyword set and the VAD emotion vector; generating an event sequence based on an emotional resonance keyword set, constructing an event graph by using the event sequence, optimizing the event sequence through a LaMDA model, calculating an edge weight, updating the event graph, obtaining probability distribution of the updated event graph through VGAE coding, and generating a plot skeleton in combination with a VAD emotional vector; and calculating a gating weight vector, calculating a dynamic user portrait vector based on the gating weight vector, calculating a loss function and updating an encoder-decoder model, and obtaining an optimized story text. The emotional consistency, the plot continuity and the individuation level of the generated stories are improved.

This invention relates to a closed-loop data system and method for embodied intelligent robots. The system includes: a data acquisition module for acquiring multimodal operational data of the embodied intelligent robot during task execution and constructing multiple embodied state nodes based on the multimodal operational data; a construction module for constructing event correlation relationships between embodied state nodes based on temporal and physical constraints, forming an embodied state event graph; an anomaly identification module for identifying anomalous embodied state nodes in the embodied state event graph and marking them as failure source nodes; and a propagation analysis module for determining failure propagation paths forward or backward from the failure source nodes based on event correlation relationships. Using the above scheme, the operational data during the task execution of the embodied intelligent robot can be effectively analyzed, anomalous states can be accurately identified, and data-driven closed-loop optimization can be achieved.

The present disclosure discloses a processing method and device of an event, a computer device and a storage medium, and the implementation scheme is as follows: a first vector corresponding to a to-be-processed event and a second vector corresponding to each reference event in a preset event graph are determined; each associated event related to the to-be-processed event is determined from the event graph according to the similarity between the first vector and the second vector; each transition probability between the to-be-processed event and each associated event is determined according to the similarity between the first vector and the second vector corresponding to each associated event; and a type label corresponding to the to-be-processed event is determined according to each transition probability and a label vector corresponding to each associated event. Thus, the type label corresponding to the to-be-processed event is determined by combining the semantic information of the to-be-processed event, each transition probability between the to-be-processed event and each associated event, and the label vector corresponding to each associated event, thereby improving the rationality and reliability of the event handling suggestion and reducing the event processing cost.

The invention discloses a logistics scheduling management method based on an ant colony algorithm. The method comprises the following steps: constructing a logistics network model constrained by a capacity time window; carrying out ant colony search to generate a scheduling scheme set and carrying out local modeling to obtain a scheduling constraint manifold; converting the scheduling scheme into a vehicle event graph and introducing a time phase to obtain an event phase code; defining a manifold projection operator on the scheduling constraint manifold, mapping an event phase code into a feasible code, and setting a water wave parameter according to a difference value; a water wave model with event phase codes as individuals is constructed, candidate codes are generated through translation, torsion and event subgraph reconnection, and projection is feasible; constructing a cost field and a risk field according to the transportation cost and the time window default risk, and updating event phase codes and ant colony parameters according to a gradient; and when the model converges, outputting a logistics scheduling scheme and generating a vehicle warehouse instruction. Scheduling constraint manifold and water wave optimization are introduced on the basis of ant colony search, and cost and risk are balanced under capacity constraint.

This application relates to the field of intelligent construction and engineering management technology, and discloses an automatic generation system for construction site engineering documents based on multimodal information acquisition. The system includes modules for on-site perception and acquisition, data transmission network, cloud-based core processing, and user interaction terminal. Based on a BIM model and construction plan, the system constructs a dynamic causal event graph. Utilizing spatiotemporal semantic anchors and multimodal projection mapping technology, it accurately associates on-site collected images and sensor data with process event nodes. A cloud-based logic verification service module calculates the completeness status of event nodes in real time. When a missing evidence chain is detected, it proactively generates reverse control commands to schedule on-site equipment to complete the data. After the node status is complete, the system automatically extracts data, generates standardized documents, and uploads them to the blockchain for storage. This invention solves the problems of semantic discrepancies in construction data, missing logical verification, and low document reliability, achieving automated closed-loop generation and reliable delivery of engineering documents.

The invention belongs to the field of signaling log analysis, and particularly relates to a dynamic self-adaptive signaling log analysis method based on a large language model, which comprises the following steps: collecting an industrial log and carrying out preprocessing operation; the self-adaptive analyzer converts the preprocessed industrial log into candidate elements and outputs confidence; constructing a physical rule of an industrial site; marking the candidate elements by using physical rules of an industrial site through a consistency calibrator; writing the coincident annotations into the event atlas; trying to deduce missing elements according to adjacent events and physical constraints by a completion engine for the non-conforming standard; judging the confidence degree of the inference result; the system periodically feeds back a verification result and a manually confirmed result to the analyzer for online fine tuning; according to the method, log analysis and credibility judgment can be completed at the edge in cooperation with the cloud end, a high-precision and low-delay dynamic self-adaptive signaling log analysis system can be achieved in a multi-source heterogeneous environment, and the automation degree of anomaly recognition and event reconstruction is remarkably improved.

This invention discloses a log anomaly detection method and apparatus for improving the security of distributed systems. The method includes: sampling structured log template files using a sliding window, mapping log events to graph nodes, and constructing a dynamic event graph that evolves over time based on the order of events; for the topological changes in the dynamic event graph, adaptively aggregating and weighting the features of neighboring nodes using a graph attention network to update the node embedding representation, and combining the residuals as the semantic embedding features of the nodes; counting the occurrence frequency of each log template within the sliding window, inputting the normalized frequency vector into a lightweight encoder composed of convolutional layers, activation functions, and pooling layers, and outputting fixed-dimensional log frequency features; concatenating the semantic embedding features obtained through the dynamic graph attention network with the log frequency features to obtain a fused feature representation; and inputting the fused feature representation into a BiGRU network based on an attention mechanism to determine whether the node is anomaly. The apparatus includes a processor and a memory.

The present disclosure provides a hotspot event mining method and device, equipment and medium, relates to the technical field of data processing, and particularly relates to the technical field of big data and artificial intelligence. The implementation scheme is: obtaining a plurality of original documents; for each original document, obtaining at least one keyword included in the original document; based on a plurality of keywords included in each of the plurality of original documents, obtaining at least one keyword frequent item set; based on the at least one keyword frequent item set, determining a plurality of preliminary screening documents from the plurality of original documents; at least based on a plurality of keywords included in each of the plurality of preliminary screening documents, constructing an event graph; based on the event graph, obtaining at least one event cluster; and based on the at least one event cluster, determining a hotspot event list.

The invention discloses a document-level event argument extraction method and system based on a semantic fusion graph, and the method comprises the steps: building a semantic mention graph by taking entity mentions in a document as nodes of the graph and semantic relationships as edges, fusing a graph structure into a large language model for embedding representation through a multi-layer encoder, and dynamically updating node and edge information through an attention mechanism. The embedded representation of the enhanced context semantics is obtained; constructing an event graph based on all event structures in the document, modeling a multi-event association relationship, and fusing the multi-event association relationship into a large language model decoder to strengthen the understanding of the model on the association between events; the method comprises the following steps: constructing cue sentences according to event types, synchronously inputting multi-event cue sentences in a training stage, designing a weighted loss function, guiding a model to learn multi-event information interaction, inputting the multi-event cue sentences in a reasoning stage, and only taking a target event argument result as an evaluation standard. According to the method, document noise can be suppressed, the problems of scattered distribution and long-distance dependence of events and arguments are solved, and the accuracy and robustness of argument extraction are improved.

The present application relates to the technical field of data collaborative processing, in particular to a cross-service financial collaborative processing method and device and storage medium. It comprises the following steps: collecting multi-source transaction and behavior event stream, and constructing global customer-event graph; implanting probe in event bus and each business processing node, and constructing influence surface blood chain; calculating customer association network characteristics and risk transmission strength characteristics; inputting two types of characteristics into collaborative rule model, outputting risk grading signal and accurate containment instruction set when characteristics match; selecting hierarchical response process template through process arrangement model, and generating workflow with minimum business interference. The present application realizes multi-source event semantic normalization through financial transaction ontology, combines probe and distributed tracking technology to connect risk information, and solves the problem that existing financial collaborative mode cannot identify and trace complex fraud transactions involving payment, credit and financial multi-business lines in real time.

The application discloses a document receiving and sending intelligent agent based on artificial intelligence and a data set generation method, and particularly relates to the field of artificial intelligence, and relates to intelligent decision-making and trainable data construction of document receiving, distribution, handling, batch handling, circulation, archiving and other businesses. The method models the evidence pointing structure, establishes a stable association between the handling conclusion and the evidence fragments in the text and the attachments, forms a traceable instruction, evidence and action unified sample structure, filters high information density samples through event graph construction and multi-view consistency anomaly analysis, and synthesizes controllable difficult cases in the form of counterfactual disturbance to improve the coverage of complex situations. In the reasoning stage, the process constraint execution mechanism is fused, the candidate actions are filtered and the rejection reason record is output, the model output is ensured to be consistent with the executable action set of the business, and the audit playback and closed-loop incremental learning are supported.

This invention discloses a campus energy consumption intelligent analysis system based on big data, belonging to the fields of big data analysis and energy management technology. It acquires multi-source energy consumption data from the campus and constructs a dynamic event graph. The system includes: a matrix construction module that builds a Bayesian causal network based on the dynamic event graph and historical energy consumption data, outputting a causal influence factor matrix; a candidate instruction module that generates a candidate instruction set based on the causal influence factor matrix using a genetic algorithm; a priority instruction module that outputs a priority instruction sequence based on the simulated energy consumption curve after executing the candidate instruction set; a collaborative optimization module that distributes the priority instruction sequence to each building node, generating device-level control instructions; and an autonomous strategy module that performs deep reinforcement learning optimization iterations based on the actual energy saving rate and comfort deviation feedback after executing the device-level control instructions. This invention effectively solves the inefficiency problems of IoT systems due to resource heterogeneity, rigid scheduling, and lack of autonomous optimization capabilities.

The invention relates to the technical field of multi-modal data processing, and discloses a live broadcast cargo carrying real-time detection method based on multi-modal fusion, and the method comprises the following steps: S1, collecting a multi-modal live broadcast data flow in real time; s2, incrementally constructing a dynamic semantic event graph based on the atomic event sequence; s3, on the dynamic semantic event graph; and S4, according to the result of the sub-graph matching detection or the structural anomaly detection. Multi-modal data streams such as video and audio texts are uniformly abstracted into structured atomic events, and a dynamic semantic event graph capable of expressing inter-event time, entity and logic association is incrementally constructed, so that deep fusion and context association of discrete and heterogeneous live broadcast information are realized; according to the method, the limitation that only shallow feature splicing can be carried out in a traditional method is overcome, internal relations and combination modes among events in different modes can be observed from a global perspective, and a uniform and semantic-rich data basis is provided for subsequent complex behavior recognition.

The application discloses a target range federation cross-domain threat judgment method and system, aiming at solving the problems of cross-domain threat correlation analysis difficulty, entity fragmentation and low judgment precision caused by the fact that original security data of multiple target ranges cannot be shared. The method comprises the following steps: each sub-target range locally performs structural processing and high-order semantic coding on the original security data to generate a standardized judgment intermediate representation; a central coordination node performs cross-domain entity disambiguation and constructs a time sequence event graph based on the intermediate representation, and outputs a global threat judgment conclusion by fusing multi-source evidence through a Bayesian network; and the result is returned to the relevant target range according to a permission policy to form a closed-loop feedback. The system comprises a judgment agent unit deployed in each sub-target range and a central coordination node, which respectively realize local feature extraction and collaborative reasoning. The application realizes cross-domain accurate restoration and collaborative defense of complex threats such as APT attack chains on the premise of guaranteeing that the original data does not go out of the domain.

The invention discloses a method and a system for automatically constructing, tracing and deducing a affair graph based on a large model, and belongs to the crossing field of artificial intelligence and knowledge engineering. The construction method comprises the following steps: respectively constructing a knowledge graph and an event graph of a target application scene based on an ontology category tree and an event category tree; transmitting the knowledge graph and the event graph to a large model to generate a affair graph; wherein the large model identifies equivalent entities in the knowledge graph and the event graph through semantic similarity calculation and performs merging, then dynamic connection between events and entities is established to obtain the affair graph, and the affair graph is composed of four-tuple basic units composed of entities, events, relations and attributes. According to the method, a traditional knowledge triple is expanded into a quadruple structure fusing knowledge and events by means of a multi-graph fusion technology, so that more reliable and interpretable strategy suggestions are provided for decision makers.

The application provides a monocular depth estimation method and product based on convolution compensation double-channel self-attention, and relates to the field of computer vision. The method comprises the following steps: converting an event stream of a target scene into a three-dimensional tensor representation; obtaining event image fusion multi-scale spatial features based on an image of the target scene and the three-dimensional tensor representation; modeling spatial context correlation of the event image fusion multi-scale spatial features in a spatial dimension by a context modeling self-attention branch to obtain a context modeling self-attention result; modeling modality correlation of the event image fusion multi-scale spatial features in a channel dimension by a modality fusion self-attention branch to obtain a modality fusion self-attention result; and performing pixel-level depth prediction by using the context modeling self-attention result and the modality fusion self-attention result to obtain a depth map, so as to fully mine the complementary characteristics between events and images, realize fine-grained depth fusion expression, and effectively improve the depth estimation precision and generalization ability.

A compute instance is managed by a threat management facility that provides security for an enterprise network associated with the compute instance, and that stores event data in a data lake for use in threat detection. In response to a security event on a compute instance, the compute instance creates a lineage for the security event that facilitates immediate presentation to a technician for review. The lineage may include data for one or more related processes so that an event graph or the like can be immediately displayed in the user interface upon receipt of the lineage. The user interface may be subsequently augmented as additional data becomes available from the data lake, or in response to requests from a user investigating the security event in the user interface.

The application discloses an event camera-based event-by-event space-time representation method. The method comprises the following steps: acquiring original event sequence data collected by an event camera; performing feature embedding on the original event sequence data in the space and time dimensions, and converting the obtained feature embedding into a fixed-dimension embedding tensor containing N tensor elements; inputting the fixed-dimension embedding tensor into a triple attention network, and calculating the feature correlation between the embedding tensor elements; constructing a multi-layer perception machine, taking the local space-time autocorrelation weight, the local space autocorrelation weight and the global space-time autocorrelation weight of each event as inputs, and calculating a complete space-time representation tensor of the event sequence. Compared with the time granularity loss caused by the common event image stacking or voxel aggregation, the application can retain all the space-time information and time granularity of the original event sequence to the greatest extent, which is of great significance for the practical application of the event camera in a high-speed scene.

The invention relates to the technical field of data processing and artificial intelligence, and discloses an intelligent repetition judgment method and system for petition repeated cases and a storage medium, and the method comprises the steps: receiving to-be-processed petition data, and generating an initial candidate appeal event graph containing uncertainty nodes; performing data interaction with an external authoritative data system to correct the uncertain nodes, and generating a final finalization appeal event graph; executing a multi-layer progressive matching process on the final finalization appeal event graph in a graph database so as to retrieve an existing graph; and when the retrieval result is low-confidence matching, starting a closed-loop confirmation process, determining a final matching relationship by positioning a mismatched hub point and initiating secondary exploration, and updating the confirmed matching knowledge to the knowledge base. According to the method, through data dynamic correction and uncertainty closed-loop confirmation, the accuracy and reliability of duplicate judgment are remarkably improved, and self-adaptive learning of the system is realized.

The invention discloses a self-learning attack process anomaly detection and processing method, which is used for explicitly converting a complex attack process into path change in a graph structure by constructing and dynamically maintaining a time sequence event graph which takes an entity as a node and takes interaction as an edge. Furthermore, multi-scale time sequence feature coding and a graph self-supervised learning technology are adopted to realize parallel capture and dynamic deviation measurement of long and short term modes of node behaviors, and finally, causal path backtracking of a reverse time sequence is performed based on abnormal anchor points, so that a complete attack chain is restored. In this way, the problem that in the prior art, long-period and slow attack detection of APT and the like is missed to report is fundamentally solved, and early discovery, accurate tracing and correlation analysis of hidden attacks are achieved.

An event query host can include an event processor configured to process an event stream indicating events that occurred on a computing device. The event processor can add representations of events to an event graph. If an event added to the event graph is a trigger event associated with a query, the event processor can also add an instance of the query to a query queue. The query queue can be sorted based on scheduled execution times of query instances. At a scheduled execution time of a query instance in the query queue, a query manager of the event query host can execute the query instance and attempt to find a corresponding pattern of one or more events in the event graph.